On Tue, Aug 01, 2006 at 11:50:38AM -0700, go moko wrote:
> 
> It's a problem linked to the -fPIC parameter which is
> not correctly set during the compilation of the tests.
> The book give a sed command to skip bad tests, but it
> is not sufficient.

 It used to be enough, at least on pure64, but I can see it in Joe's
results for x86_64, so point taken.

> What I suggest you is to use the following command,
> after the configure command of the book (and without
> the sed before in the book), to force the -fPIC
> option:
> sed -i 's/gcc -m64/gcc -m64 -fPIC/g'
> autoconf-lib-link/tests/Makefile
> 
> On my machine, with that, all the tests passed without
> any error (of course, if you want, you can also put
> -fPIC in your CFLAGS).
> 
> I put the clfs-dev list in copy, because I think it's
> perhaps better to replace the sed in the book but this
> one, or a better on before configure, if possible.
>
 A quick look in autoconf-lib-link/ suggests that @CFLAGS@ are picked
up during configure, so if we are going to do it more cleanly we
will either pass CC="gcc ${BUILD64} -fPIC" or CFLAGS="-fPIC" to
configure.  Either of these will probably cause the whole compile to
use -fPIC, I hope that is not a problem, but it needs to be tested.
I guess CFLAGS is a safer idea, if it works.

 Anybody able to test any of this ?

> What do you think of that?
> 
> G. Moko
> 
 I think you should get yourself an account at trac.cross-lfs.org
and open a ticket for this.  You might also want to ticket your
earlier comments about the perl test failures when booting without
'ip'.

Regards,

Ken
-- 
das eine Mal als Tragödie, das andere Mal als Farce
_______________________________________________
Clfs-dev mailing list
[email protected]
http://lists.cross-lfs.org/cgi-bin/mailman/listinfo/clfs-dev

Reply via email to